Skip to content

Advertisement

You're viewing the new version of our site. Please leave us feedback.

Learn more

BMC Neuroscience

Open Access

Best practices for avoiding dominant experimental bias in analysis of multielectrode array signals

  • Tyler Stone1,
  • Uzair Khawaja1,
  • Nardeen Perko1,
  • Thomas R Kiehl2Email author and
  • Charles Bergeron1
BMC Neuroscience201415(Suppl 1):P208

https://doi.org/10.1186/1471-2202-15-S1-P208

Published: 21 July 2014

Multielectrode arrays (MEA) are broadly used for in vitro cell culture observation. Complete realization of this platform’s potential to the study of disease and development depends on the capability to make 1) reliable and repeatable observations, 2) neuronal network behavior characterizations, and 3) meaningful comparisons between cultures. Using an extensive, previously published, set of recordings [3] we demonstrate that naïve processing of this data leaves us vulnerable to bias from measurement error.

We utilize 878 recordings from embryonic rat cortex cell cultures collected from 60-electrode, grid-type (200 μm) MEA’s. We modeled each recording as a 60-node directed weighted graph with weights describing electrode connectivity and nodal clustering coefficients [1] as features. Principal components analysis (PCA) reduced the dimensionality of this feature space [2]. We observe a dense core of recordings showing weak spike activity, and 3 classes defined by batch (labeled in Figure 1A). Surprised at this structure, we sought a non-biological explanation. We identified 13 defective or biased electrodes as sources of systematic measurement error. Removing the affected electrodes produced a more complex interplay of inter- and intra-batch variability (Figure 1B).
Figure 1

(A) Principal components analysis biplots using all 60 electrodes. Each marker is a recording. Colors indicate batches: Batch 1 (purple), Batch 2 (red), Batch 3(orange), Batch 4 (light green), Batch 5 (grey), Batch 6 (blue), Batch 7 (pink), and Batch 8 (dark green). The large black circle is the dense core of recordings showing weak spike activity. Classes A, B and C are also labeled. (B) PCA biplots using the 47 electrodes following deletion of biased ones. Some regions are associated with specific cultures identified by culture numbers and batch colors.

Bias from defective electrodes can be avoided wholly by proper documentation of experimental conditions. This highlights the need for best practices when recording MEA signals. Our analysis highlights possibilities for post-experimental review to identify unknown issues and retroactively handle them. Thereupon, we make a case for transparency in data reporting and propose best practices for experimental and analysis phases.

Authors’ Affiliations

(1)
Analytics Lab, Albany College of Pharmacy and Health Sciences
(2)
Neural Stem Cell Institute

References

  1. Bullmore E, Sporns O: Complex brain networks: graph theoretical analysis of structural and functional systems. Nat Rev Neurosci. 2009, 10: 186-198.View ArticlePubMedGoogle Scholar
  2. Jolliffe IT: Principal Component Analysis,. 2002, New York: Springer, SecondGoogle Scholar
  3. Wagenaar DA, Pine J, Potter SM: An extremely rich repertoire of bursting patterns during the development of cortical cultures. BMC Neurosci. 2006, 7 (11):Google Scholar

Copyright

© Stone et al; licensee BioMed Central Ltd. 2014

This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

Advertisement